Meta-analysis: New Tests for the Diagnosis of Latent TuberculosisInfection: Areas of Uncertainty and Recommendations for ResearchDick Menzies, MD, MSc; Madhukar Pai, MD, PhD; and George Comstock, MD, DrPH
Background: Until recently, the tuberculin skin test was the onlytest for detecting latent tuberculosis (TB) infection, but 2 ex vivointerferon-� release assays (IGRAs) are now commercially licensed.
Purpose: To estimate sensitivity, specificity, and reproducibility ofIGRAs (commercial or research versions of QuantiFERON [QFT]and Elispot) for diagnosing latent TB infection in healthy and im-mune-suppressed persons.
Data Sources: The authors searched MEDLINE and reviewed cita-tions of all original articles and reviews for studies published inEnglish.
Study Selection: Studies had evaluated IGRAs using Mycobacte-rium tuberculosis–specific antigens (RD1 antigens) and overnight(16- to 24-h) incubation times. Reference standards had to beclearly defined without knowledge of test results.
Data Extraction and Quality Assessment: Specific criteria forquality assessment were developed for sensitivity, specificity, andreproducibility.
Data Synthesis: When newly diagnosed active TB was used as asurrogate for latent TB infection, sensitivity of all tests was subop-timal, although it was higher with Elispot. No test distinguishesactive TB from latent TB. Sensitivity of the tuberculin skin test andIGRAs was similar in persons who were categorized into clinical
gradients of exposure. Pooled specificity was 97.7% (95% CI, 96%to 99%) and 92.5% (CI, 86% to 99%) for QFT and for Elispot,respectively. Both assays were more specific than the tuberculin skintest in samples vaccinated with bacille Calmette–Guerin. Elispot wasmore sensitive than the tuberculin skin test in 3 studies of immune-compromised samples. Discordant tuberculin skin test and IGRAreactions were frequent and largely unexplained, although somemay be related to varied definitions of positive test results. Rever-sion of IGRA results from positive to negative was common in 2studies in which it was assessed.
Limitations: Most studies used cross-sectional designs with theinherent limitation of no gold standard for latent TB infection, andmost involved small samples with a widely varying likelihood oftrue-positive and false-positive test results. There is insufficient ev-idence on IGRA performance in children, immune-compromisedpersons, and the elderly.
Conclusions: New IGRAs show considerable promise and haveexcellent specificity. Additional studies are needed to better definetheir performance in high-risk populations and in serial testing.Longitudinal studies are needed to define the predictive value ofIGRAs.
Ann Intern Med. 2007;146:340-354. www.annals.orgFor author affiliations, see end of text.
The tuberculin skin test is 1 of the few tests that hasbeen in use for nearly 100 years in clinical medicine
(1). Therefore, it is not surprising that the test has impor-tant limitations. The tuberculin skin test uses a relativelycrude mix of antigens from Mycobacterium tuberculosis. Asa result, false-positive reactions can occur because of previ-ous bacille Calmette–Guerin (BCG) vaccination or sensi-tization to nontuberculous mycobacteria. False-negative re-sults on tuberculin skin tests can occur because of severeillness, including active tuberculosis (TB), or immune sup-pression, often due to HIV infection. Initial testing canaffect results of subsequent tests because of anamnestic re-call of immunity (the booster effect). Errors in administra-tion or reading can lead to incorrect results. The variabilityof the tuberculin skin test can be minimal with well-trained personnel using meticulous techniques (2–8), al-though such personnel are not always available. Interpreta-tion of test results requires a thorough understanding ofthe test.
In the past decade, 2 new T-cell–based tests for diag-nosing latent TB infection have been developed and li-censed for commercial distribution in many countries (Ap-pendix Table 1, available at www.annals.org). One test,QuantiFERON (QFT)-TB Gold (Cellestis, Victoria, Aus-tralia), uses enzyme-linked immunosorbent assay to mea-
sure antigen-specific production of interferon-� (IFN-�)by circulating T cells in whole blood. The other test, theT-SPOT.TB (Oxford Immunotec, Oxford, United King-dom), uses the Elispot technique to measure peripheralblood mononuclear cells that produce IFN-�. Current ver-sions of both IFN-� release assays (IGRAs) use more spe-cific M. tuberculosis antigens—ESAT-6, CFP-10, andTB7.7. The genes encoding these antigens are found in theregion of difference 1 (RD1) of the M. tuberculosis genome,which is deleted from the genome of M. bovis BCG, andcertain nontuberculous mycobacteria, such as M. avium.The U.S. Centers for Disease Control and Prevention (9)has recommended that QFT replace the tuberculin skin
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
test, whereas the U.K. National Institute for Clinical Ex-cellence has suggested that IGRAs are useful adjuncts tothe tuberculin skin test (10).
Interferon �–release assays have several important ad-vantages over the tuberculin skin test: Testing requires only1 patient visit and these assays are ex vivo tests, whichreduce the risk for adverse effects and eliminate potentialboosting when testing is repeated. However, IGRAs haveimportant disadvantages, including higher material cost,need for an equipped laboratory, and a requirement todraw blood with subsequent careful handling to maintainviability of lymphocytes. Although boosting will not occur,the variability of these tests when repeated after months oryears, such as in serial testing of exposed populations, hasnot been well studied. The greatest disadvantage is the lackof prospective data regarding the future risk for active TBin persons with positive results on IGRAs. This has beenestablished for different-sized tuberculin skin test reactionsin many large-scale cohort and experimental studies (11–20), which permits the estimation of risk for disease andbenefit of therapy.
In this review, we compared sensitivity, specificity, andreproducibility of commercially available IGRAs (or in-house equivalent tests) with the tuberculin skin test fordiagnosing latent TB infection in healthy and immune-suppressed adults and children. Because there is no goldstandard for testing latent TB infection, we estimated sen-sitivity from studies of 1) patients with active TB (whomust have infection), 2) persons in contact with patientswith active TB who were categorized into gradients of ex-posure, and 3) concordance of IGRA and the tuberculinskin test. We estimated specificity from studies of healthypersons with a very low likelihood of exposure. Reproduc-ibility was estimated from studies of repeated IGRAs, withor without treatment for active or latent TB.
METHODS
Search StrategyWe conducted a MEDLINE search for articles pub-
lished between 1966 and October 2006 on 4 December2005 and updated the search on 2 July 2006 and 31 October2006. Search terms included TB infection, or TB disease,AND Quantiferon, or Elispot, or interferon-gamma assays,or interferon-gamma release assays, or T cell assays, ANDESAT-6, or CFP10, or RD1 antigens. The searches werelimited to human studies published in English.
We identified additional studies from the referencelists of articles identified in this manner and performed ahand search of all issues of the International Journal ofTuberculosis and Lung Disease that were published over thepast 5 years. We contacted the authors of some studies forclarification.
Study SelectionStudies were included if they used QFT or Elispot
tests with overnight (16 to 24 h) in vitro incubation of
peripheral blood lymphocytes stimulated with RD1 anti-gens. This meant research, in-house, or commercial ver-sions of QFT or Elispot tests that used ESAT-6 with orwithout CFP-10 and with or without TB7.7. Commercialversions that were included were the QuantiFERON-TBGold (QFT-G), QuantiFERON-TB Gold In-Tube (QFT-IT), and the T-SPOT.TB.
For studies assessing sensitivity, the study sample hadto have active TB or contact with a person with active TBand a clearly defined gradient of exposure of at least 2categories (such as high or low exposure). For studies as-sessing specificity, the population had to be healthy, life-long residents of low-incidence countries with an averageage younger than 40 years without occupational, travel, orother exposure to TB. In studies assessing concordance of 2or more tests, all tests had to be done simultaneously in allpersons. We excluded studies that performed sequentialtesting, in which the second test was done only in personsselected on the basis of the results of the first test. Studiesassessing reproducibility had to perform at least 2 tests inthe same persons. Two independent reviewers assessed ti-tles and abstracts to select studies for the review. A list ofexcluded studies is available from the authors.
Extraction of Data and Assessment of Study QualityOne reviewer reviewed all studies and abstracted data
regarding test characteristics and study quality (AppendixTables 2 through 6, available at www.annals.org). A sec-ond reviewer assessed a random sample of 20% of thefull-text articles to determine concordance in assessment ofdata quality and accuracy of abstracted data. We assessed 2quality criteria in all studies: The technicians performing
Context
Do new ex vivo interferon-� release assays (IGRAs) detectlatent tuberculosis (TB) more accurately than tuberculinskin tests (TSTs)?
Contribution
This meta-analysis of 59 studies found that no test distin-guished active TB from latent TB, no test had high sensi-tivity, IGRAs were more specific than the TST in popula-tions vaccinated with bacille Calmette–Guerin (BCG), andthe results of IGRAs and the TST were frequently discor-dant.
Cautions
Studies had many limitations, including no gold standardfor diagnosing latent TB.
Implications
New IGRAs have good specificity and show promise fordetecting latent TB, particularly in BCG-vaccinated pa-tients.
—The Editors
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 341
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Figure 1. Forest plot of studies estimating sensitivity of the 3 tests in patients with active tuberculosis as a surrogate for latenttuberculosis infection.
Point estimates for sensitivity and 95% CIs are shown. QFT � QuantiFERON; TST � tuberculin skin test.
Article Diagnosis of Latent Tuberculosis Infection
342 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
the tuberculin skin test and IGRAs were blinded to thepatients’ clinical status, other test results and timing ofIGRA relative to the tuberculin skin test, and treatment (ifapplicable). In all studies estimating sensitivity, the clinicalgold standard had to be determined without knowing orusing results of the tests that were being evaluated. In ad-dition, high-quality studies of active TB (as a surrogate forlatent TB infection) confirmed the diagnosis microbiolog-ically or histologically. High-quality studies of reproduc-ibility had high rates of participation of eligible persons,and high rates of completion of testing protocols and as-sessed treatment adherence or outcomes in studies oftreated persons.
Data Synthesis and AnalysisMethods of studies estimating sensitivity among pa-
tients with active TB as a surrogate for latent TB infectionand estimating specificity in low-risk populations were sim-
ilar enough to allow pooled estimates. In particular, thesame diagnostic thresholds were used for in-house assaysand commercial assays. Different cut-points were used forthe tuberculin skin test; these were analyzed separately.
For each study, we calculated sensitivity or specificity(and 95% CIs) and summarized the results in forest plots.A fixed-effects meta-analysis was done by using Meta-DiSc,version 1.2 (Ramon y Cajal Hospital, Madrid, Spain) (21):Studies were weighted by total sample size to pool (sum-marize) estimates of sensitivity and specificity across thestudies. Presence of statistically significant heterogeneityacross studies was evaluated by using the chi-square test forheterogeneity. To account for heterogeneity due to be-tween-study variability, we corrected for overdispersion.
We did not pool estimates of test sensitivity on thebasis of gradients of exposure, concordance among high-risk persons, and effects of immune suppression or treat-
Figure 2. Forest plot of studies estimating specificity of the 3 tests in populations at very low risk for latent tuberculosis infection.
Point estimates for specificity and 95% CIs are shown. BCG � bacille Calmette–Guerin; QFT � QuantiFERON; TST � tuberculin skin test.
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 343
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
ment, because of the heterogeneity of ascertainment andclassification of exposure and study samples.
Role of the Funding SourceNo funding sources directly supported the study. Dr.
Menzies receives salary support from the Fonds de laRecherche en Sante du Quebec, which had no role in theconduct of or in the decision to publish the study.
RESULTS
Studies IdentifiedA total of 275 studies were identified from PubMed,
of which 83 were selected for full-text review (AppendixFigure, available at www.annals.org). From these originalarticles and reviews, 21 additional articles were identifiedand 6 more were identified at the time of the last update,for a total of 110 studies for full review. Of these studies,52 were excluded because they did not meet the reviewcriteria, leaving 58 studies included in this review. All stud-ies used cross-sectional designs, except for 8 that performedserial testing in cohorts. Studies ranged from fewer than 10participants to more than 500 participants. The expectedprevalence of latent TB infection and the prevalence andpolicy of BCG vaccination also varied widely.
Twenty-two studies evaluated sensitivity of IGRAamong newly diagnosed patients with active TB as a sur-rogate for latent TB infection. Of these studies, 3 testedboth IGRAs and the tuberculin skin test, 12 tested 1 IGRAand the tuberculin skin test, and 7 tested 1 IGRA only.Specificity of IGRA in populations at very low risk forlatent TB infection was assessed in 11 studies, of which 6performed the tuberculin skin test. Ten studies compared asingle IGRA with the tuberculin skin test in contacts witha gradient of exposure, and 14 studies examined concor-dance of IGRA with the tuberculin skin test. Serial testingwas performed in 7 treated cohorts and in 1 additionaluntreated cohort. Five studies examined the impact of im-mune suppression on IGRA performance; 3 of these com-pared results with those of the tuberculin skin test.
Sensitivity Using Tuberculosis as a Surrogate forLatent Infection
Figure 1 shows that all 3 tests, particularly the tuber-culin skin test, have suboptimal sensitivity among patientswith newly diagnosed, active TB. Pooled estimates of sen-sitivity were lowest for the tuberculin skin test, higher forQFT, and highest for Elispot. However, only 3 studies,with a total of 143 participants, conducted direct head-to-head comparisons (22–24). There were some differences insensitivity between studies using different versions of QFT.QuantiFERON with ESAT-6 alone was used only once inan older study with earlier assay and antigens. This studyreported substantially lower sensitivity but higher specific-ity (Figure 2). Sensitivity of QFT-G was somewhat higherthan with QFT-IT (Tables 1 and 2). Sensitivities of Eli-spot assays that incorporated ESAT-6 alone were similar tothose of assays with ESAT-6 plus CFP-10. The IGRA testprocedures were well standardized, but tuberculin skin testprocedures were not. As shown in Table 1, different cut-points were used, and this had some effect on sensitivity. In4 studies, which used a dose of tuberculin that was 50% orless than the recommended amount (26–28, 85), sensitiv-ity of the tuberculin skin test averaged 0.63 (range, 0.46 to0.80) compared with a sensitivity of 0.73 (range, 0.62 to0.84) in studies that used a tuberculin dose recommendedby the World Health Organization (2).
Sensitivity Using a Gradient of Exposure as an Indicatorof the Likelihood of Latent Tuberculosis Infection
Table 3 summarizes studies that compared IGRA andthe tuberculin skin test with a clinically defined gradient ofexposure. It is difficult to ensure equivalence of the expo-sure categories because each study characterized exposuredifferently. Nevertheless, overall findings were similar. Theprevalence of positive results on IGRA and the tuberculinskin test was highest in the most-exposed groups. In theless-exposed groups, the prevalence of positive results onthe tuberculin skin test was higher than that of IGRA instudies that involved populations that had received BCGvaccination at an older age (30, 31). In 1 study, the ex-pected prevalence of latent TB infection was 30% in the
Table 1. Summary of Sensitivity from Pooled Estimates fromAll Studies*
* Patients with active tuberculosis were used as surrogates for latent tuberculosis.NS � not significant.† All 95% CIs are corrected for overdispersion.
Article Diagnosis of Latent Tuberculosis Infection
344 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
least-exposed group, yet only 4% had a positive result onIGRA (30).
Specificity of TestsIn all studies of healthy populations considered at very
low risk for latent TB infection, IGRA with RD1 antigenshave had high specificity (Figure 2). Pooled average speci-ficity was 97.7% and 92.2% in QFT and Elispot, respec-tively (Tables 1 and 2), and was unaffected by BCG vac-cination. A limitation of these studies is that persons at lowrisk could still have had latent TB infection, but this couldnot be verified because of the absence of a true gold stan-dard for testing for latent TB infection.
Discordance between Interferon-� Release Assays andTuberculin Skin Test: Reflection of Uncertainty onSensitivity and Specificity of These Tests
Tables 4 and 5 summarize studies of dual tuberculinskin testing and IGRA testing of populations considered atrisk for TB infection. In 3 studies, discordance was greaterin persons with BCG vaccination than in those who werenot vaccinated (30, 32, 33). In 2 studies in which very fewpersons had been BCG vaccinated, almost half of all per-sons with positive results on tuberculin skin test had neg-ative results on IGRA (34, 35). Discordant positive IGRAreactions and negative tuberculin skin test reactions oc-curred in 6% to 7% of all persons and accounted for 23%of all positive results on QFT and 26% of all positiveresults on Elispot. In 3 studies summarized in Table 6,agreement between QFT-G and Elispot was good: � valueswere 0.57 to 0.70. Of note, Elispot results were more oftenpositive than those of QFT, which is consistent withhigher sensitivity or lower specificity as noted earlier. In a
fourth study (80), QFT-G was compared with QFT-IT.Concordance was moderate (� � 0.5): QFT-IT had agreater prevalence of positive reactions. The differences inthese 2 versions of the same test may reflect poor repro-ducibility under field conditions, higher sensitivity ofQFT-IT because of the addition of the third antigen(TB7.7), or that QFT-IT was more sensitive because of itsgreater technical simplicity.
Serial TestingVery few published studies have assessed reproducibil-
ity of IGRAs. In studies conducted by the manufacturer,the test-related coefficient of variation for the QFT was8.7% by using replicate serum samples from well-charac-terized patients (37). Among healthy volunteers fromGambia in whom Elispot was repeated 1 week apart, 12initially had negative results and none had conversion topositive results, but of the 11 who initially had positiveresults, 1 (9%) reverted to having negative results (38). Ina study of health care workers in India who were tested 18months apart, 11.6% had QFT conversion, but 24% ofthose with initial positive results on QFT reverted to hav-ing negative results (39). QuantiFERON reversion was as-sociated with negative results on the initial tuberculin skintest or an initial QFT result close to the manufacturer’ssuggested cut-point for a positive result.
As seen in Table 7, in 4 studies of patients treated foractive TB, serial Elispot assays decreased in 3 studies (38,40, 41), increased after 1 month in a fourth study, thendecreased with continued therapy (42). In patients whowere treated for latent TB infection, IGRA results did notchange in 2 studies (26, 43) but reverted to negative results
Table 2. Summary of Specificity from Pooled Estimates from Studies of Persons at Very Low Risk for Tuberculosis Infection*
Elispot or T-SPOT.TBAll studies 4 0.92 (0.88–0.95) 21.3 0.01
* BCG � bacille Calmette–Guerin; NS � not significant.† In 1 study (30), data for 2 tuberculin skin test cut-points are given.‡ In each of 2 studies (68, 69), 2 different very-low-risk populations were tested. These were counted as separate studies.
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 345
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
in 16% of treated persons in the third study (44). Ofimportance, in this last study, 7 of 25 (28%) persons whowere not treated had reversion—a phenomenon seen inthose who initially had positive results on Elispot and neg-ative results on tuberculin skin test (44).
Impact of Immune SuppressionTo date, few studies have assessed IGRA in these high-
risk populations. In 2 studies (45, 46), responses of IGRAwere slightly reduced in immune-compromised patientscompared with those who were not immune-compro-mised. In 3 studies with direct comparison, the prevalenceof positive results on Elispot was substantially higher thanthat of positive results on the tuberculin skin test (46–48),particularly in persons with greater immune suppression(47). The only study to evaluate QFT noted a low preva-lence of positive results in asymptomatic HIV-infected pa-tients and that indeterminate results were more common ifthe patient’s CD4 count was less than 100 cells/mm3 (49).
Results in Pediatric PopulationsStudies of IGRA response in pediatric samples are
more heterogeneous because the populations selected weresuspected of having active TB (23, 26, 50), had contactwith infected persons (23, 36, 51), or were school children
(52). Some were healthy (36, 52, 53), whereas others werehospitalized (26, 50); some studies were conducted in low-income countries (26, 36, 51, 52), and others were con-ducted in middle-income (53) or high-income (23, 50)countries. Confirming or excluding active TB is also moredifficult in children. Hence, information is currently insuf-ficient to estimate sensitivity, specificity, and reproducibil-ity of IGRAs in children, although the occurrence of inde-terminate QFT (23, 50) and failed phlebotomy (51, 52)may be important potential limitations of IGRAs in thispopulation.
DISCUSSION
Principal FindingsCommercially available IGRAs have evolved rapidly
over the past decade. The latest versions use more specificM. tuberculosis antigens and are simpler to perform. Thesetests are promising and have been successfully used by in-dependent investigators in many settings, including low-income countries. In our review, the sensitivity of the tu-berculin skin test compared with QFT, using active TB asa surrogate for latent TB infection, was low but similar forboth tests, whereas the sensitivity of Elispot using this sur-
Table 3. Studies Comparing Interferon-� Release Assay (IGRA) with Tuberculin Skin Test (TST) against the Clinical Gold Standardof Exposure Gradients*
Study, Year(Reference)
Country Patients with a Historyof BCG vaccination, n(%)
Antigens for IGRA-Positiveand TST-Positive Type
Results by Exposure Gradient†
High Exposure
Participantsin ExposureCategory, n
IGRA-Positive,%
TST-Positive,%
ElispotLalvani et al.,
2001 (31)United Kingdom 59 (most) ESAT-6
HEAF (PPD)20 73 65
Ewer et al.,2003 (41)
United Kingdom 535 (most) ESAT and CFP-10HEAF (PPD)
20 100 90
Richeldi et al.,2004 (70)
Italy 92 (10) ESAT-6 and CFP-10TST: 5TU-PPD
4 25 0
Hill et al.,2004 (71)
Gambia 735 (45) ESAT-6 and CFP-10TST: 2TU-RT23
149 38 62
Zellweger et al.,2005 (72)
Switzerland 91 (86) ESAT-6 and CFP-10TST: 2TU-RT23
54 22 50
Shams et al.,2005 (73)
United States 413 (49} ESAT-6 and CFP-10TST: 5TU-PPDS
104 32 28
Hill et al.,2006 (36)
Gambia 718 (46) ESAT-6 and CFP-10TST: 2TU-RT23
163 46 53
QuantiFERONBrock et al.,
2004 (34)Denmark 85 (0) ESAT-6 and CFP-10
TST: 2TU-RT2345 46 53
Kang et al.,2005 (30)
Korea 273 (81) ESAT-6 and CFP-10TST: 2TU-RT23
48 44 71
Nakaoka et al.,2006 (51)
Nigeria 207 (90) ESAT-6 and CFP-10TST: 5TU-PPDS
72 74 53
* BCG � bacille Calmette–Guerin; PPD � purified protein derivative; PPDS � purified protein derivative standard.† The percentage is the proportion in that exposure category with a positive response to IGRA or TST.‡ This group represented a random sample from the general population. On the basis of historical data, the authors estimated that the prevalence of latent tuberculosisinfection should have been 30%. On the basis of the average age of 25 years, the incidence of smear-positive tuberculosis (40/100 000) (74), and the Styblo equation (75),the estimated prevalence of latent tuberculosis infection should have been 18.5%.
Article Diagnosis of Latent Tuberculosis Infection
346 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
rogate was somewhat superior. When a gradient of expo-sure among contacts was used as the gold standard, sensi-tivity of IGRA and the tuberculin skin test were similar.None of these tests can distinguish between latent andactive TB.
Specificity of IGRA is excellent because neither IGRAis affected by BCG vaccination, although the effect of non-tuberculous mycobacteria on IGRA response is poorlystudied. However, BCG vaccination has a greater effect onthe specificity of the tuberculin skin test, particularly ifBCG was given after infancy. There is substantial discor-dance of IGRA with the tuberculin skin test in populationswith varying likelihood of latent TB infection. Althoughsome discordance may be explained by superior specificityof IGRA, it would be overly simplistic to assume that theresults of IGRA tests were always correct and that those oftuberculin skin test were always incorrect. Some discor-dance may be explained by 1 test being close to the criteriafor positive test results, but this phenomenon has not beenanalyzed in most studies and thus is not well-understood.
In 3 studies of immune-compromised patients, sensi-tivity of Elispot was superior to that of the tuberculin skintest. Studies in pediatric populations were too heteroge-neous to draw firm conclusions. Studies of the effect of
treatment on IGRA response are contradictory. However,the true effect cannot be ascertained until the biologicaland random variability of IGRA response has been bettercharacterized. In 2 studies, a substantial number of personswith positive results on IGRA reverted to having negativeresults without treatment. This was associated with initialnegative results on the tuberculosis skin test (that is, dis-cordant results) (39, 44) and IGRA response close to thecut-point for positive test results (39).
LimitationsThe major limitation of estimation of sensitivity or
specificity was the use of a cross-sectional design in allstudies reviewed. With this design, there is no gold stan-dard for latent TB infection. The only certain measure thatlatent TB infection exists is when the risk for active TBassociated with a particular test result has been defined.This requires large-scale cohort studies with long-term fol-low-up of untreated populations with positive results atbaseline. In addition to being expensive and complex, suchstudies are ethically impossible in most high-income coun-tries, where the standard of care is to offer treatment tosuch persons.
With cross-sectional designs, surrogates for true infec-
Table 3—Continued
Results by Exposure Gradient†
Moderate Exposure Low Exposure Very Low Exposure
Participantsin ExposureCategory, n
IGRA-Positive,%
TST-Positive,%
Participantsin ExposureCategory, n
IGRA-Positive,%
TST-Positive,%
Participantsin ExposureCategory, n
IGRA-Positive,%
TST-Positive,%
8 38 43 12 0 31 7 0 33
81 53 51 47 38 40 387 17 20
9 67 11 16 6 6 63 14 3
340 30 41 246 24 28 – – –
– – – 37 5 35 – – –
103 28 26 103 21 25 103 19 20
372 33 32 182 20 15 – – –
– – – 40 5 10 – – –
72 10 60 99 4‡ 51‡ – – –
81 10 16 39 10 15 – – –
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 347
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
tion must be used. The most commonly used proxy isactive TB, because persons with disease must be infected.However, with active TB, the cell-mediated immune re-sponse is often diminished, particularly at the time of di-agnosis, in patients with more advanced disease (54–56),malnutrition (57), or older age (55, 58). Intuitively, onewould expect reduced performance of tests, such as thetuberculin skin test or IGRA, because the cell-mediatedimmunity they measure must have failed, to some extent,in any person with active disease. Of interest, the greatersensitivity of Elispot in active TB is paralleled by findingsin immune-suppressed patients. We speculate that this mayreflect the technical requirement for standardization of thenumber of peripheral blood mononuclear cells in each as-say well.
An additional important limitation is the heterogene-ity of the studies reviewed, which were conducted withdifferent timing of testing and in different settings withsamples that had markedly different risks for exposure and
infection. Because of these methodological differences,pooling of estimates could be made only for results of sen-sitivity among patients with active TB (as a surrogate) andof specificity in low-risk populations. The estimation of thesensitivity of these tests among contacts who were catego-rized into gradients of exposure was limited by the differ-ences in degree and categorization of exposure. Thus, dif-ferences in results may have reflected differences inexposure or exposure definition rather than differences intest performance. There were too few studies that evaluatedIGRAs in immune-compromised persons and in pediatricpatients to make firm conclusions.
To date, most studies of the new IGRA have includedthe investigators who developed these assays. This is inev-itable because of their recent introduction, but could resultin 2 problems. First, highly complex tests could achieveinitially excellent results in research laboratories, but testperformance may be reduced in the hands of independentinvestigators who lack the specialized expertise to overcome
Table 4. Concordance of QuantiFERON and Tuberculin Skin Test (TST) in Healthy Populations with Varying Risk for LatentTuberculosis Infection*
Study, Year(Reference)
Country Risk Group TotalParticipants,n
BCG, % Age at BCGVaccination
Concordant Results Discordant Results
TST-PositiveandIGRA-Positive,n (%)
TST-
andIGRA-Negative,n (%)
TST-NegativeandIGRA-Positive,n (%)
TST-PositiveandIGRA-Negative,n (%)
Brock et al.,2004 (34)
Denmark Contacts ofpersons withtuberculosis
45 0 23 (51) 19 (42) 1 (2) 2 (4)
Pai et al.,2005 (76)
India Health careworkers
719 71 Infancy 223 (31) 360 (50) 65 (9) 72 (10)
Kang et al.,2005 (30)
Korea Close and casualcontacts ofpersons withtuberculosis
120 73 Older 24 (20) 40 (33) 4 (3) 53 (44)
Porsa et al.,2006 (35)
U.S. Prisoners 409 All ages† 8 (2) 360 (88) 12 (3) 29 (7)
* BCG � bacille Calmette–Guerin; IGRA � interferon-� release assay.† In these 3 studies, the participants who had received BCG vaccination were immigrants from many countries in which BCG vaccination policy differed.‡ Study participants had previous multiple BCG vaccinations and TSTs.§ Eight participants with active tuberculosis were excluded from this analysis.� Results from QuantiFERON-Gold In-Tube (ESAT-6/CFP-10/TB7.7) are shown. Results with QuantiFERON-Gold (ESAT-6/CFP-10) were less concordant with TST.
Article Diagnosis of Latent Tuberculosis Infection
348 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Negative
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
technical challenges. Second, investigators could have aconflict of interest if they retained financial interest in theassays.
The strengths of this review include the focus on com-mercially available versions of the new IGRAs or in-houseversions that used identical techniques, making the resultsmore relevant to practicing clinicians. In addition, most ofthe studies we included were published in the past 2 years,making this review as up-to-date as possible. However, be-cause this is an active field of investigation, estimated sen-sitivity, specificity, and reproducibility may change withadditional publications, particularly where evidence islimited.
ImplicationsThe most consistent finding in this review is the high
specificity of IGRA, which was unaffected by BCG vacci-nation, in all populations. This reflects the antigens used inthese assays, which are not present in BCG or in certainnontuberculous mycobacteria (59). In the studies reviewed,the effect of BCG vaccination on the results of the tuber-culin skin test varied, which may reflect differences in pa-tients’ ages when the vaccination was administered. In anearlier review of 24 studies that involved 240 243 personswho had been vaccinated with BCG in infancy and a sim-ilar number of persons who were not vaccinated, false-
positive results on tuberculin skin tests (�10 mm) attrib-utable to BCG vaccination occurred in 6.3% personsoverall and only in 1% of those who were tested after morethan 10 years (60). In the same review, in the 12 studiesinvolving 12 728 persons who were vaccinated with BCGat 2 years of age or older, BCG was responsible for false-positive results on tuberculin skin tests in 40% of all per-sons and in 20% after 10 or more years (60). It is clear thatIGRA will be most useful in BCG-vaccinated populations,particularly in those that received BCG vaccination repeat-edly or after infancy. This practice was common in Europeand Latin America for many years, but now global vacci-nation policy is to vaccinate infants only (61). In popula-tions that received BCG vaccinations as infants, the IGRAwill be slightly more specific in childhood but no morespecific in adolescence or in adult life (60).
None of the reviewed studies assessed the effect ofnontuberculous mycobacteria on IGRA response. Personswith disease caused by M. marinum and M. kansasei canhave positive results on RD1-IGRA (62), which is of in-terest because the genomes of these nontuberculous myco-bacteria organisms contain the RD1 region (59). It wouldbe interesting to perform RD1 on the basis of IGRA anddual skin testing with purified protein derivative (or RT23)plus nontuberculous mycobacteria antigens, such as M.
Table 5. Concordance of Elispot or T-SPOT.TB with Tuberculin Skin Test (TST) in Healthy Populations with Varying Risk for LatentTuberculosis Infection*
* BCG � bacille Calmette–Guerin; IGRA � interferon-� release assay.† The participants who had received BCG vaccination were immigrants from many countries where BCG vaccination policy differed.
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 349
Negative
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
marinum, M. kansasei, M. szulgai, or M. flavescens, fromorganisms containing the RD1 region. In our earlier re-view, we found that for every 100 persons with nontuber-culous mycobacteria sensitivity, only 2 (2%) would havefalse-positive results, measuring 10 to 14 mm, on tubercu-lin skin tests. This effect was reasonably consistent in allcountries where dual testing was performed, despite thewide variation in prevalence in nontuberculous mycobac-teria sensitivity (60). This means that even in populationsin which every person was sensitized to nontuberculousmycobacteria, tuberculin skin test specificity would still be98%. This modest effect will be clinically important onlyin populations with a high prevalence of nontuberculousmycobacterial sensitivity and a very low prevalence of trueTB infection (such as in the southern United States [63,64]).
For the diagnosis of latent TB infection, the sensitivityof IGRA is less clear, although in study samples with agradient of exposure, the prevalence of positive test resultswas similar in the most-exposed categories. The biggestproblem in estimating sensitivity is the lack of a propergold standard for latent TB infection. In cross-sectionalstudies, the most commonly used surrogate of latent TBinfection has been newly diagnosed, active TB. However,this is a poor surrogate because of the known reduction incell-mediated response in such patients. The ideal goldstandard in cross-sectional studies would be healthy per-sons known to have TB infection. The only patients whomeet these criteria are those being treated for active TB,who have clinically recovered. In an older multicenterstudy of more than 1000 such patients, tuberculin skin testreactions were remarkably similar in all populations, andoverall sensitivity exceeded 93% (29). In 3 earlier studies ofsuch patients, sensitivity of purified protein derivative–based IGRA ranged from 59% to 71%; however, tubercu-lin skin test sensitivity was 95% (65). Does this mean that
the IGRA is less sensitive or that treatment changes theimmune response to the IGRA measure? The results of ourreview indicate that IGRA response generally decreasedwith treatment in patients who had active TB, althoughthis phenomenon was not seen in patients with latent TBinfection. But, why and how would patients with activeTB differ from those with untreated latent TB infectionafter 3 to 6 months of treatment?
This issue is unresolved. Unfortunately, there are norecent studies using RD1-based IGRA in patients whohave recovered from active TB, although these would be ofgreat interest. The effect of treatment on IGRA responsealso cannot be ascertained without understanding the in-herent biological variability in IGRA response over monthsor years. Short-term variability was approximately 9% in 2studies (37, 38). However, in the only 2 studies with fol-low-up of 1 to 2 years, spontaneous reversion of IGRA wasdocumented in 24% to 28% of persons and was associatedwith discordant and weaker initial IGRA responses (39,44). Until the variability of IGRA response has been stud-ied, the effect of treatment on IGRA response cannot beascertained. Without this information, the accuracy ofevaluation of IGRA sensitivity will not be known in pa-tients with active TB who have recovered clinically.
The numerous studies that have used active TB as asurrogate for latent infection may give rise to the miscon-ception that these tests are useful for diagnosing active TB.However, they do not have optimal sensitivity, and moreimportant, cannot distinguish active from latent TB, whichseverely reduces their specificity for active TB. The onlyexception is diagnosis of active TB in children, becausetuberculin skin testing is commonly used for this indica-tion. Results from our review indicate that sensitivity ofQFT and Elispot are somewhat lower in children than inadults. There are too few studies to make definitive state-ments, but the results of this review support the urgent
Table 6. Concordance of T-SPOT.TB and QuantiFERON*
Study, Year(Reference)
Country Risk Group Participants,n
BCGVaccination,%
Age at BCGVaccination
Concordant Reactions Discordant Reactions
T-SPOT.TB–PositiveandQFT-G–Positive,n (%)
T-SPOT.TB–
andQFT-G–Negative,n (%)
T-SPOT.TB–PositiveandQFT-G–Negative,n (%)
T-SPOT.TB–NegativeandQFT-G–Positive,n (%)
Ferrara et al.,2006 (23)
Italy Persons withsuspected TB
383 18 All ages 94 (25) 233 (61) 50 (13) 6 (2)
Lee et al.,2006 (22)
Korea Persons withactive TBand those atlow risk forTB
Article Diagnosis of Latent Tuberculosis Infection
350 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Negative
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
need for further evaluation of IGRA in the diagnosis of TBin pediatric patients.
We hope that in the near future, the gold standard willbe longitudinal studies that determine the incidence of ac-tive TB in persons with positive and negative results onIGRA and tuberculin skin tests. Results from these studieswill establish the risk for TB in persons with positive re-sults on IGRA—particularly those with discordant nega-tive results on tuberculin skin testing. As noted in ourreview, these discordant reactions are common and poorlyunderstood at present, rendering clinical management ofpersons with such reactions very difficult. Some discor-dance may be explained by random variation with 1 orboth tests being close to the cut-point for positivity (66,67).
One important use of tuberculin skin testing is serialtesting of exposed populations to detect new TB infection.The variability of tuberculin skin testing has been wellestablished, but the variation in IGRA response has not.
Two studies have shown that similar to the tuberculin skintest, IGRA shows nonspecific variations and reversionsfrom positive to negative results during serial testing (39,44). Until information is available to define IGRA con-version, results of serial IGRA testing will be largelyuninterpretable.
Recommendations for ResearchOn the basis of the data presented, we recommend
further research in the following areas. Independent studiesof the reproducibility of IGRA should be done. To esti-mate test variability, repeated assays should be performedon the same samples by the same technician, by differenttechnicians in the same laboratory, or in different labora-tories. The effect of minor modifications in the field testingprotocol, such as time from phlebotomy to incubation,time of incubation, or temperature of storage, should betested in the same way. To estimate biological variability,repeated tests should be repeated in unexposed and un-
Table 7. Effect of Treatment on Interferon-� Release Assay (IGRA) Response in All Cohort Studies Using RD1 Antigens andOvernight Assays Only*
Study, Year(Reference)
Country Test Type(Incubation Time)
Participants Days WhenTested
Change orDifference†
Details
Pathan et al.,2001 (40)
United Kingdom Elispot (14 h) Persons with active TB:12 tested before,during, and aftertreatment
0 and30–240
Decrease 62% had a decrease inmean levels over anaverage of 19 wk
Carrara et al.,2004 (78)
Italy Elispot Persons with active TB:18 were treated
0, 90, and180
Decrease 13 of 18 had reversionto negative results
Nicol et al.,2005 (42)
South Africa Elispot (18 h) Pediatric patients withactive TB: 15 hadprobable or possibledisease
0 and 30 Increase thendecrease
Mean levels increasedby 45% after 1 moof therapy comparedwith before therapy
Aiken et al.,2006 (38)
Gambia Elispot (6–14 h) Persons with active TB:82 tested before andafter treatment
0 and 365 Decrease 82% had positiveresults beforetreatment, and 46%had positive results6 mo after treatment
Ewer et al.,2006 (44)
United Kingdom Elispot Persons with latent TBinfection: 38 withpositive TST andIGRA results weretreated; 11 withpositive TST andIGRA results werenot treated; 14 withnegative TST resultsand positive IGRAresults were nottreated
0, 180, 365,and 640
Decrease, nochange, anddecrease
6 of 38 (16%) hadreversion aftertreatment; 0 of11 (0%) hadreversion who werenot treated; and 7of 14 (50%) hadreversion withouttreatment
Wilkinson et al.,2006 (79)
United Kingdom Elispot (14 h) Persons with latent TBinfection: 33 receivedINH/RIF; 8 receivedno treatment
0, 26, and82
No change Mean levels increasedduring treatment butdecreased at the endof treatment; nochange was seen inuntreated persons
Pai et al.,2006 (43)
India QFT-IT (16–20 h) Persons with latent TBinfection: 10 receivedINH
0, 365, and640
No change Median levels 10 U 35 U 3 7.9 U;percentage positive(�0.35): 100% 390% 3 90%
* In this study, differences between group means were less marked and less significant after 18 hours than after 6 days incubation, but trends were similar. INH � isoniazid; RIF �rifampin; QFT-IT � QuantiFERON-Gold In-Tube; TB � tuberculosis; TST � tuberculin skin test.† Change or difference comparing results on or after treatment relative to before treatment.
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 351
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
treated persons at intervals ranging from days to years. Thisinformation is crucial to distinguish random variation fromconversions attributable to new TB infection and to studythe effect of treatment on IGRA response.
More studies are needed that compare the sensitivityof the tuberculin skin test with IGRA in HIV-infected andother immune-compromised groups, intravenous drug us-ers, and pediatric and elderly populations.
More data are needed to understand discordant tuber-culin skin test and IGRA reactions, including the effect ofchanges in cut-points, the role of nontuberculous myco-bacteria, and time to conversion after exposure andinfection.
Now that both IGRAs are widely commercially avail-able, independent field studies can evaluate the feasibility,utility, and costs of these tests in different populations andunder different conditions. Such studies should report onthe actual completion of tests and the subsequent evalua-tion and treatment for patients with positive test results.Much of the value of a test is lost if persons who are testeddo not return to learn the significance of their test result orif those with positive results are not evaluated further.With these results, the cost-effectiveness of IGRA and thetuberculin skin test can be compared in real-world settings.
Finally, large-scale cohort studies are needed that esti-mate risk for progression to active disease in persons whohave had tuberculin skin test and IGRA. Of particularinterest is the risk for disease in persons with discordantreactions.
From McGill University, Montreal, Quebec, Canada, and Johns Hop-kins University, Baltimore, Maryland.
Acknowledgments: The authors thank Drs. Janice Pogoda, Peter Barnes,Philip Hill, Thomas Meier, Peter Andersen, and Delia Goletti for pro-viding additional information.
Grant Support: None.
Potential Financial Conflicts of Interest: None disclosed.
Requests for Single Reprints: Dick Menzies, MD, MSc, RespiratoryEpidemiology and Clinical Research Unit, Montreal Chest Institute,3650 St-Urbain, Room K1.24, Montreal, Quebec H2X 2P4, Canada;e-mail, [email protected].
Current author addresses and author contributions are available at www.annals.org.
References1. Mantoux MC. La voie intradermique en tuberculinotherapie. Presse Med.1912;20:146-48.2. World Health Organization. The WHO standard tuberculin test. Geneva:World Health Organization; 1963.3. Bearman JE, Kleinman H, Glyer VV, Lacroix OM. A study of variablilty intuberculin test reading. Am Rev Respir Dis. 1964;90:913-9. [PMID: 14233795]4. Perez-Stable EJ, Slutkin G. A demonstration of lack of variability among sixtuberculin skin test readers. Am J Public Health. 1985;75:1341-3. [PMID:
4051078]5. Erdtmann FJ, Dixon KE, Llewellyn CH. Skin testing for tuberculosis. Anti-gen and observer variability. JAMA. 1974;228:479-81. [PMID: 4594690]6. Pouchot J, Grasland A, Collet C, Coste J, Esdaile JM, Vinceneux P. Reli-ability of tuberculin skin test measurement. Ann Intern Med. 1997;126:210-4.[PMID: 9027271]7. Furcolow ML, Watson KA, Charron T, Lowe J. A comparison of the tine andMono-Vacc tests with the intradermal tuberculin test. Am Rev Respir Dis. 1967;96:1009-27. [PMID: 6059186]8. Chaparas SD, Vandiviere HM, Melvin I, Koch G, Becker C. Tuberculin test.Variability with the Mantoux procedure. Am Rev Respir Dis. 1985;132:175-7.[PMID: 4014862]9. Mazurek GH, Jereb J, Lobue P, Iademarco MF, Metchock B, Vernon A.Guidelines for using the QuantiFERON-TB Gold test for detecting Mycobacte-rium tuberculosis infection, United States. MMWR Recomm Rep. 2005;54:49-55. [PMID: 16357824]10. Royal College of Physicians. Tuberculosis: national clinical guidelines fordiagnosis, management, prevention, and control. London: Royal College of Phy-sicians; 2006. Accessed at www.nice.org.uk.11. Comstock GW, Edwards LB, Livesay VT. Tuberculosis morbidity in theU.S. Navy: its distribution and decline. Am Rev Respir Dis. 1974;110:572-80.[PMID: 4429253]12. Comstock GW, Livesay VT, Woolpert SF. The prognosis of a positivetuberculin reaction in childhood and adolescence. Am J Epidemiol. 1974;99:131-8. [PMID: 4810628]13. Ferebee SH. Controlled chemoprophylaxis trials in tuberculosis. A generalreview. Bibl Tuberc. 1970;26:28-106. [PMID: 4903501]14. Efficacy of various durations of isoniazid preventive therapy for tuberculosis:five years of follow-up in the IUAT trial. International Union Against Tubercu-losis Committee on Prophylaxis. Bull World Health Organ. 1982;60:555-64.[PMID: 6754120]15. Nolan CM, Elarth AM. Tuberculosis in a cohort of Southeast Asian Refu-gees. A five-year surveillance study. Am Rev Respir Dis. 1988;137:805-9.[PMID: 3354985]16. Wood R, Maartens G, Lombard CJ. Risk factors for developing tuberculosisin HIV-1-infected adults from communities with a low or very high incidence oftuberculosis. J Acquir Immune Defic Syndr. 2000;23:75-80. [PMID: 10708059]17. Selwyn PA, Hartel D, Lewis VA, Schoenbaum EE, Vermund SH, KleinRS, et al. A prospective study of the risk of tuberculosis among intravenous drugusers with human immunodeficiency virus infection. N Engl J Med. 1989;320:545-50. [PMID: 2915665]18. Grzybowski S. Ontario studies on tuberculin sensitivity. Can J Public Health.1965;56:181-92. [PMID: 14285275]19. Falk A, Fuchs GF. Prophylaxis with isoniazid in inactive tuberculosis. AVeterans Administration Cooperative Study XII. Chest. 1978;73:44-8. [PMID:340155]20. Hawken MP, Meme HK, Elliott LC, Chakaya JM, Morris JS, Githui WA,et al. Isoniazid preventive therapy for tuberculosis in HIV-1-infected adults: re-sults of a randomized controlled trial. AIDS. 1997;11:875-82. [PMID: 9189212]21. Zamora J, Abraira V, Muriel A, Khan K, Coomarasamy A. Meta-DiSc: asoftware for meta-analysis of test accuracy data. BMC Med Res Methodol. 2006;6:31. [PMID: 16836745]22. Lee JY, Choi HJ, Park IN, Hong SB, Oh YM, Lim CM, et al. Comparisonof two commercial interferon-gamma assays for diagnosing Mycobacterium tuber-culosis infection. Eur Respir J. 2006;28:24-30. [PMID: 16611658]23. Ferrara G, Losi M, D’Amico R, Roversi P, Piro R, Meacci M, et al. Use inroutine clinical practice of two commercial blood tests for diagnosis of infectionwith Mycobacterium tuberculosis: a prospective study. Lancet. 2006;367:1328-34.[PMID: 16631911]24. Goletti D, Carrara S, Vincenti D, Saltini C, Rizzi EB, Schinina V, et al.Accuracy of an immune diagnostic assay based on RD1 selected epitopes foractive tuberculosis in a clinical setting: a pilot study. Clin Microbiol Infect. 2006;12:544-50. [PMID: 16700703]25. An Q, Buxton D, Hendricks A, Robinson L, Shah J, Lu L, et al. Compar-ison of amplified Q beta replicase and PCR assays for detection of Mycobacteriumtuberculosis. J Clin Microbiol. 1995;33:860-7. [PMID: 7540627]26. Dogra S, Narang P, Mendiratta DK, Chaturvedi P, Reingold AL, ColfordJM Jr, et al. Comparison of a whole blood interferon-gamma assay with tuber-culin skin testing for the detection of tuberculosis infection in hospitalized chil-dren in rural India. J Infect. 2006. DOI: 10.1016/J.Jinf.2006.04.007 (electronic
Article Diagnosis of Latent Tuberculosis Infection
352 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
access). [PMID: 16733068]27. Lalvani A, Pathan AA, McShane H, Wilkinson RJ, Latif M, Conlon CP, etal. Rapid detection of Mycobacterium tuberculosis infection by enumeration ofantigen-specific T cells. Am J Respir Crit Care Med. 2001;163:824-8. [PMID:11282752]28. Lalvani A, Nagvenkar P, Udwadia Z, Pathan AA, Wilkinson KA, ShastriJS, et al. Enumeration of T cells specific for RD1-encoded antigens suggests ahigh prevalence of latent Mycobacterium tuberculosis infection in healthy urbanIndians. J Infect Dis. 2001;183:469-77. [PMID: 11133379]29. World Health Organization, Tuberculosis Research Centre. Further studiesof geographic variation in naturally acquired tuberculin sensitivity. Bull WorldHealth. 1955;22:63-83.30. Kang YA, Lee HW, Yoon HI, Cho B, Han SK, Shim YS, et al. Discrepancybetween the tuberculin skin test and the whole-blood interferon gamma assay forthe diagnosis of latent tuberculosis infection in an intermediate tuberculosis-bur-den country. JAMA. 2005;293:2756-61. [PMID: 15941805]31. Lalvani A, Pathan AA, Durkan H, Wilkinson KA, Whelan A, Deeks JJ, etal. Enhanced contact tracing and spatial tracking of Mycobacterium tuberculosisinfection by enumeration of antigen-specific T cells. Lancet. 2001;357:2017-21.[PMID: 11438135]32. Harada N, Nakajima Y, Higuchi K, Sekiya Y, Rothel J, Mori T. Screeningfor tuberculosis infection using whole-blood interferon-gamma and Mantouxtesting among Japanese healthcare workers. Infect Control Hosp Epidemiol.2006;27:442-8. [PMID: 16671023]33. Diel R, Nienhaus A, Lange C, Meywald-Walter K, Forssbohm M,Schaberg T. Tuberculosis contact investigation with a new, specific blood test ina low-incidence population containing a high proportion of BCG-vaccinatedpersons. Respir Res. 2006;7:77. [PMID: 16707012]34. Brock I, Weldingh K, Lillebaek T, Follmann F, Andersen P. Comparison oftuberculin skin test and new specific blood test in tuberculosis contacts. Am JRespir Crit Care Med. 2004;170:65-9. [PMID: 15087297]35. Porsa E, Cheng L, Seale MM, Delclos GL, Ma X, Reich R, et al. Compar-ison of a new ESAT-6/CFP-10 peptide-based gamma interferon assay and atuberculin skin test for tuberculosis screening in a moderate-risk population. ClinVaccine Immunol. 2006;13:53-8. [PMID: 16426000]36. Hill PC, Brookes RH, Adetifa IM, Fox A, Jackson-Sillah D, Lugos MD, etal. Comparison of enzyme-linked immunospot assay and tuberculin skin test inhealthy children exposed to Mycobacterium tuberculosis. Pediatrics. 2006;117:1542-8. [PMID: 16651307]37. Cellestis. QuantiFERON Precision and reproducibility report. Accessed atwww.cellestis.com/IRM/content/gold/Precisionproductivity.pdf on 27 February27 2006.38. Aiken AM, Hill PC, Fox A, McAdam KP, Jackson-Sillah D, Lugos MD, etal. Reversion of the ELISPOT test after treatment in Gambian tuberculosis cases.BMC Infect Dis. 2006;6:66. [PMID: 16573826]39. Pai M, Joshi R, Dogra S, Mendiratta DK, Narang P, Kalantri S, et al. Serialtesting of health care workers for tuberculosis using interferon-gamma assay. AmJ Respir Crit Care Med. 2006;174:349-55. [PMID: 16690977]40. Pathan AA, Wilkinson KA, Klenerman P, McShane H, Davidson RN,Pasvol G, et al. Direct ex vivo analysis of antigen-specific IFN-gamma-secretingCD4 T cells in Mycobacterium tuberculosis-infected individuals: associations withclinical disease state and effect of treatment. J Immunol. 2001;167:5217-25.[PMID: 11673535]41. Ewer K, Deeks J, Alvarez L, Bryant G, Waller S, Andersen P, et al. Com-parison of T-cell-based assay with tuberculin skin test for diagnosis of Mycobac-terium tuberculosis infection in a school tuberculosis outbreak. Lancet. 2003;361:1168-73. [PMID: 12686038]42. Nicol MP, Pienaar D, Wood K, Eley B, Wilkinson RJ, Henderson H, et al.Enzyme-linked immunospot assay responses to early secretory antigenic target 6,culture filtrate protein 10, and purified protein derivative among children withtuberculosis: implications for diagnosis and monitoring of therapy. Clin InfectDis. 2005;40:1301-8. [PMID: 15825033]43. Pai M, Joshi R, Dogra S, Mendiratta DK, Narang P, Dheda K, et al.Persistently elevated T cell interferon-gamma responses after treatment for latenttuberculosis infection among health care workers in India: a preliminary report. JOccup Med Toxicol. 2006;1:7. [PMID: 16722616]44. Ewer K, Millington KA, Deeks JJ, Alvarez L, Bryant G, Lalvani A. Dy-namic antigen-specific T-cell responses after point-source exposure to Mycobacte-rium tuberculosis. Am J Respir Crit Care Med. 2006;174:831-9. [PMID:16799072]
45. Liebeschuetz S, Bamber S, Ewer K, Deeks J, Pathan AA, Lalvani A. Diag-nosis of tuberculosis in South African children with a T-cell-based assay: a pro-spective cohort study. Lancet. 2004;364:2196-203. [PMID: 15610806]46. Chapman AL, Munkanta M, Wilkinson KA, Pathan AA, Ewer K, Ayles H,et al. Rapid detection of active and latent tuberculosis infection in HIV-positiveindividuals by enumeration of Mycobacterium tuberculosis-specific T cells.AIDS. 2002;16:2285-93. [PMID: 12441800]47. Piana F, Codecasa LR, Cavallerio P, Ferrarese M, Migliori GB, BarbaranoL, et al. Use of a T-cell-based test for detection of tuberculosis infection amongimmunocompromised patients. Eur Respir J. 2006;28:31-4. [PMID: 16540502]48. Passalent L, Khan K, Richardson R, Wang J, Deider H, Gardam M.Detecting latent tuberculosis infection in hemodialysis patients: a head-to-headcomparison of the T-SPOT.TB test, tuberculin skin test, and expert physicianpanel. Clin J Am Soc Nephrol. 2006;1-6.49. Brock I, Ruhwald M, Lundgren B, Westh H, Mathiesen LR, Ravn P.Latent tuberculosis in HIV positive, diagnosed by the M. tuberculosis specificinterferon-gamma test. Respir Res. 2006;7:56. [PMID: 16579856]50. Connell TG, Curtis N, Ranganathan SC, Buttery JP. Performance of awhole blood interferon gamma assay for detecting latent infection with Mycobac-terium tuberculosis in children. Thorax. 2006;61:616-20. [PMID: 16601088]51. Nakaoka H, Lawson L, Squire SB, Coulter B, Ravn P, Brock I, et al. Riskfor tuberculosis among children. Emerg Infect Dis. 2006;12:1383-8. [PMID:17073087]52. Tsiouris SJ, Austin J, Toro P, Coetzee D, Weyer K, Stein Z, et al. Resultsof a tuberculosis-specific IFN-gamma assay in children at high risk for tubercu-losis infection. Int J Tuberc Lung Dis. 2006;10:939-41. [PMID: 16898381]53. Soysal A, Millington KA, Bakir M, Dosanjh D, Aslan Y, Deeks JJ, et al.Effect of BCG vaccination on risk of Mycobacterium tuberculosis infection inchildren with household tuberculosis contact: a prospective community-basedstudy. Lancet. 2005;366:1443-51. [PMID: 16243089]54. Holden M, Dubin MR, Diamond PH. Frequency of negative intermediate-strength tuberculin sensitivity in patients with active tuberculosis. N Engl J Med.1971;285:1506-9. [PMID: 5127141]55. Battershill JH. Cutaneous testing in the elderly patient with tuberculosis.Chest. 1980;77:188-9. [PMID: 7353412]56. Howard WL, Klopfenstein MD, Steininger WJ, Woodruff CE. The loss oftuberculin sensitivity in certain patients with active pulmonary tuberculosis.Chest. 1970;57:530-4. [PMID: 4994745]57. Kardjito T, Donosepoetro M, Grange JM. The Mantoux test in tuberculo-sis: correlations between the diameters of the dermal responses and the serumprotein levels. Tubercle. 1981;62:31-5. [PMID: 7268918]58. Stead WW, To T. The significance of the tuberculin skin test in elderlypersons. Ann Intern Med. 1987;107:837-42. [PMID: 3688676]59. Andersen P, Munk ME, Pollock JM, Doherty TM. Specific immune-baseddiagnosis of tuberculosis. Lancet. 2000;356:1099-104. [PMID: 11009160]60. Farhat M, Greenaway C, Pai M, Menzies D. False-positive tuberculin skintests: what is the absolute effect of BCG and non-tuberculous mycobacteria? IntJ Tuberc Lung Dis. 2006;10:1192-204. [PMID: 17131776]61. World Health Organization. Global BCG vaccination coverage 2000-2005.Accessesed at www.who.int/immunization_monitoring/en/globalsummary/timeseries/tscoveragebcg.htm on 24 December 2005.62. Arend SM, van Meijgaarden KE, de Boer K, de Palou EC, van SoolingenD, Ottenhoff TH, et al. Tuberculin skin testing and in vitro T cell responses toESAT-6 and culture filtrate protein 10 after infection with Mycobacterium mari-num or M. kansasii. J Infect Dis. 2002;186:1797-807. [PMID: 12447766]63. Palmer CE, Edwards LB. Identifying the tuberculous infected. The dual-testtechnique. JAMA. 1968;205:167-9. [PMID: 5694785]64. Edwards LB, Acquiviva FA, Livesay VT, Palmer CE. An atlas of sensitivityto tuberculin and PPD-B in the United States. Am Rev Resp Dis. 1969;99:1-132.65. Pai M, Riley LW, Colford JM Jr. Interferon-gamma assays in the immuno-diagnosis of tuberculosis: a systematic review. Lancet Infect Dis. 2004;4:761-76.[PMID: 15567126]66. Richeldi L, Ewer K, Losi M, Hansell DM, Roversi P, Fabbri LM, et al.Early diagnosis of subclinical multidrug-resistant tuberculosis. Ann Intern Med.2004;140:709-13. [PMID: 15126254]67. Pai M, Kalantri S, Menzies D. Discordance between tuberculin skin test andinterferon-gamma assays [Letter]. Int J Tuberc Lung Dis. 2006;10:942-3.[PMID: 16898382]68. Brock I, Munk ME, Kok-Jensen A, Andersen P. Performance of whole
ArticleDiagnosis of Latent Tuberculosis Infection
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 353
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
blood IFN-gamma test for tuberculosis diagnosis based on PPD or the specificantigens ESAT-6 and CFP-10. Int J Tuberc Lung Dis. 2001;5:462-7. [PMID:11336278]69. Johnson PD, Stuart RL, Grayson ML, Olden D, Clancy A, Ravn P, et al.Tuberculin-purified protein derivative-, MPT-64-, and ESAT-6-stimulatedgamma interferon responses in medical students before and after Mycobacteriumbovis BCG vaccination and in patients with tuberculosis. Clin Diagn Lab Immu-nol. 1999;6:934-7. [PMID: 10548589]70. Richeldi L, Ewer K, Losi M, Bergamini BM, Roversi P, Deeks J, et al.T cell-based tracking of multidrug resistant tuberculosis infection after brief ex-posure. Am J Respir Crit Care Med. 2004;170:288-95. [PMID: 15130907]71. Hill PC, Brookes RH, Fox A, Fielding K, Jeffries DJ, Jackson-Sillah D, etal. Large-scale evaluation of enzyme-linked immunospot assay and skin test fordiagnosis of Mycobacterium tuberculosis infection against a gradient of exposure inThe Gambia. Clin Infect Dis. 2004;38:966-73. [PMID: 15034828]72. Zellweger JP, Zellweger A, Ansermet S, de Senarclens B, Wrighton-SmithP. Contact tracing using a new T-cell-based test: better correlation with tubercu-losis exposure than the tuberculin skin test. Int J Tuberc Lung Dis. 2005;9:1242-7. [PMID: 16333932]73. Shams H, Weis SE, Klucar P, Lalvani A, Moonan PK, Pogoda JM, et al.Enzyme-linked immunospot and tuberculin skin testing to detect latent tubercu-losis infection. Am J Respir Crit Care Med. 2005;172:1161-8. [PMID:16081545]74. TB. A global emergency, WHO, July 1994. Lepr Rev. 1995;66:270-1.[PMID: 7500828]75. Styblo K. The relationship between the risk of tuberculosis infection and therisk of developing infectious tuberculosis. Bull Int Union Tuberc. 1985;60:117-19.76. Pai M, Gokhale K, Joshi R, Dogra S, Kalantri S, Mendiratta DK, et al.Mycobacterium tuberculosis infection in health care workers in rural India: com-parison of a whole-blood interferon gamma assay with tuberculin skin testing.JAMA. 2005;293:2746-55. [PMID: 15941804]77. Meier T, Eulenbruch HP, Wrighton-Smith P, Enders G, Regnath T. Sen-sitivity of a new commercial enzyme-linked immunospot assay (T SPOT-TB) fordiagnosis of tuberculosis in clinical practice. Eur J Clin Microbiol Infect Dis.2005;24:529-36. [PMID: 16133410]78. Carrara S, Vincenti D, Petrosillo N, Amicosante M, Girardi E, Goletti D.Use of a T cell-based assay for monitoring efficacy of antituberculosis therapy.Clin Infect Dis. 2004;38:754-6. [PMID: 14986262]
79. Wilkinson KA, Kon OM, Newton SM, Meintjes G, Davidson RN, PasvolG, et al. Effect of treatment of latent tuberculosis infection on the T cell responseto Mycobacterium tuberculosis antigens. J Infect Dis. 2006;193:354-9. [PMID:16388482]80. Mahomed H, Hughes EJ, Hawkridge T, Minnies D, Simon E, Little F, etal. Comparison of mantoux skin test with three generations of a whole bloodIFN-gamma assay for tuberculosis infection. Int J Tuberc Lung Dis. 2006;10:310-6. [PMID: 16562712]81. Scholvinck E, Wilkinson KA, Whelan AO, Martineau AR, Levin M,Wilkinson RJ. Gamma interferon-based immunodiagnosis of tuberculosis: com-parison between whole-blood and enzyme-linked immunospot methods. J ClinMicrobiol. 2004;42:829-31. [PMID: 14766863]82. Dewan PK, Grinsdale J, Liska S, Wong E, Fallstad R, Kawamura LM.Feasibility, acceptability, and cost of tuberculosis testing by whole-blood inter-feron-gamma assay. BMC Infect Dis. 2006;6:47. [PMID: 16539718]83. Ravn P, Munk ME, Andersen AB, Lundgren B, Nielsen LN, Lillebaek T,et al. Reactivation of tuberculosis during immunosuppressive treatment in a pa-tient with a positive QuantiFERON-RD1 test. Scand J Infect Dis. 2004;36:499-501. [PMID: 15307581]84. Ferrara G, Losi M, Meacci M, Meccugni B, Piro R, Roversi P, et al.Routine hospital use of a new commercial whole blood interferon-gamma assayfor the diagnosis of tuberculosis infection. Am J Respir Crit Care Med. 2005;172:631-5. [PMID: 15961696]85. Mori T, Sakatani M, Yamagishi F, Takashima T, Kawabe Y, Nagao K, etal. Specific detection of tuberculosis infection: an interferon-gamma-based assayusing new antigens. Am J Respir Crit Care Med. 2004;170:59-64. [PMID:15059788]86. Taggart EW, Hill HR, Ruegner RG, Litwin CM. Evaluation of an in vitroassay for interferon gamma production in response to the Mycobacterium tuber-culosis-synthesized peptide antigens ESAT-6 and CFP-10 and the PPD skin test.Am J Clin Pathol. 2006;125:467-73. [PMID: 16613353]87. Pai M, Joshi R, Dogra S, Mendiratta DK, Narang P, Kalantri S, et al.Preliminary results from an evaluation of a whole blood interferon-gamma assayin individuals with active tuberculosis in a rural Indian hospital. Infection. 2007;[In press].88. Dewan PK, Grinsdale J, Kawamura LM. Low sensitivity of a whole-bloodinterferon-gamma releaae assay for detection of active tuberculosis. Clin InfectDis. 2007;44:69-73. [PMID: 17143818]
Article Diagnosis of Latent Tuberculosis Infection
354 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Current Author Addresses: Dr. Menzies: Respiratory Epidemiology andClinical Research Unit, Montreal Chest Institute, 3650 St-Urbain,Room K1.24, Montreal, Quebec H2X 2P4, Canada.Dr. Pai: Deptartment of Epidemiology, Biostatistics & OccupationalHealth, McGill University, 1020 Pine Avenue West, Montreal, QuebecH3A 1A2, Canada.Dr. Comstock: Johns Hopkins Training Center for Public Health Re-search, PO Box 2067, 1302 Pennsylvania Avenue, Hagerstown, MD21742-3230.
Author Contributions: Conception and design: D. Menzies, M. Pai.Analysis and interpretation of the data: D. Menzies, M. Pai, G. Com-stock.Drafting of the article: D. Menzies.Critical revision of the article for important intellectual content: M. Pai,G. Comstock.Final approval of the article: M. Pai, G. Comstock.Statistical expertise: D. Menzies, M. Pai.Collection and assembly of data: D. Menzies, M. Pai.
A1. Dewan PK, Grinsdale J, Liska S, Wong E, Fallstad R, Kawamura LM.Feasibility, acceptability, and cost of tuberculosis testing by whole-blood interfer-on-gamma assay. BMC Infect Dis. 2006;6:47. [PMID: 16539718]A2. Khan K, Muennig P, Behta M, Zivin JG. Global drug-resistance patternsand the management of latent tuberculosis infection in immigrants to the UnitedStates. N Engl J Med. 2002;347:1850-9. [PMID: 12466510]A3. Oxlade O, Schwartzman K, Menzies D. Interferon-gamma release assaysand TB screening in high income countries: A cost effectiveness analysis. Int JTuberc Lung Dis. 2006.A4. Johnson PD, Stuart RL, Grayson ML, Olden D, Clancy A, Ravn P, et al.Tuberculin-purified protein derivative-, MPT-64-, and ESAT-6-stimulatedgamma interferon responses in medical students before and after Mycobacteriumbovis BCG vaccination and in patients with tuberculosis. Clin Diagn Lab Immu-nol. 1999;6:934-7. [PMID: 10548589]A5. Brock I, Munk ME, Kok-Jensen A, Andersen P. Performance of wholeblood IFN-gamma test for tuberculosis diagnosis based on PPD or the specificantigens ESAT-6 and CFP-10. Int J Tuberc Lung Dis. 2001;5:462-7. [PMID:11336278]A6. Mori T, Sakatani M, Yamagishi F, Takashima T, Kawabe Y, Nagao K, etal. Specific detection of tuberculosis infection: an interferon-gamma-based assayusing new antigens. Am J Respir Crit Care Med. 2004;170:59-64. [PMID:15059788]A7. Kang YA, Lee HW, Yoon HI, Cho B, Han SK, Shim YS, et al. Discrepancybetween the tuberculin skin test and the whole-blood interferon gamma assay forthe diagnosis of latent tuberculosis infection in an intermediate tuberculosis-burden country. JAMA. 2005;293:2756-61. [PMID: 15941805]A8. Ravn P, Munk ME, Andersen AB, Lundgren B, Nielsen LN, Lillebaek T,et al. Reactivation of tuberculosis during immunosuppressive treatment in a pa-tient with a positive QuantiFERON-RD1 test. Scand J Infect Dis. 2004;36:499-501. [PMID: 15307581]A9. Ferrara G, Losi M, Meacci M, Meccugni B, Piro R, Roversi P, et al.Routine hospital use of a new commercial whole blood interferon-gamma assayfor the diagnosis of tuberculosis infection. Am J Respir Crit Care Med. 2005;172:631-5. [PMID: 15961696]A10. Connell TG, Curtis N, Ranganathan SC, Buttery JP. Performanceof a whole blood interferon gamma assay for detecting latent infection withMycobacterium tuberculosis in children. Thorax. 2006;61:616-20. [PMID:16601088]A11. Ferrara G, Losi M, D’Amico R, Roversi P, Piro R, Meacci M, et al. Usein routine clinical practice of two commercial blood tests for diagnosis of infec-tion with Mycobacterium tuberculosis: a prospective study. Lancet. 2006;367:1328-34. [PMID: 16631911]A12. Lee JY, Choi HJ, Park IN, Hong SB, Oh YM, Lim CM, et al. Compar-ison of two commercial interferon-gamma assays for diagnosing Mycobacteriumtuberculosis infection. Eur Respir J. 2006;28:24-30. [PMID: 16611658]A13. Goletti D, Carrara S, Vincenti D, Saltini C, Rizzi EB, Schinina V, et al.Accuracy of an immune diagnostic assay based on RD1 selected epitopes foractive tuberculosis in a clinical setting: a pilot study. Clin Microbiol Infect. 2006;
12:544-50. [PMID: 16700703]A14. Dogra S, Narang P, Mendiratta DK, Chaturvedi P, Reingold AL, ColfordJM Jr, et al. Comparison of a whole blood interferon-gamma assay with tuber-culin skin testing for the detection of tuberculosis infection in hospitalized chil-dren in rural India. J Infect; 2006. DOI: 10.1016/J.Jinf.2006.04.007 (electronicaccess). [PMID: 16733068]A15. Pai M, Joshi R, Dogra S, Mendiratta DK, Narang P, Kalantri S, et al.Preliminary results from an evaluation of a whole blood interferon-gamma assayin individuals with active tuberculosis in a rural Indian hospital. Int J Tuber LungDis. 2005.A16. Dewan PK, Grinsdale J, Kawamura LM. Low sensitivity of a whole-bloodinterferon-gamma release assay for detection of active tuberculosis. Clin InfectDis. 2007;44:69-73. [PMID: 17143818]A17. Lalvani A, Pathan AA, McShane H, Wilkinson RJ, Latif M, Conlon CP,et al. Rapid detection of Mycobacterium tuberculosis infection by enumeration ofantigen-specific T cells. Am J Respir Crit Care Med. 2001;163:824-8. [PMID:11282752]A18. Pathan AA, Wilkinson KA, Klenerman P, McShane H, Davidson RN,Pasvol G, et al. Direct ex vivo analysis of antigen-specific IFN-gamma-secretingCD4 T cells in Mycobacterium tuberculosis-infected individuals: associations withclinical disease state and effect of treatment. J Immunol. 2001;167:5217-25.[PMID: 11673535]A19. Lalvani A, Nagvenkar P, Udwadia Z, Pathan AA, Wilkinson KA, ShastriJS, et al. Enumeration of T cells specific for RD1-encoded antigens suggests ahigh prevalence of latent Mycobacterium tuberculosis infection in healthy urbanIndians. J Infect Dis. 2001;183:469-77. [PMID: 11133379]A20. Chapman AL, Munkanta M, Wilkinson KA, Pathan AA, Ewer K, AylesH, et al. Rapid detection of active and latent tuberculosis infection in HIV-positive individuals by enumeration of Mycobacterium tuberculosis-specificT cells. AIDS. 2002;16:2285-93. [PMID: 12441800]A21. Scholvinck E, Wilkinson KA, Whelan AO, Martineau AR, Levin M,Wilkinson RJ. Gamma interferon-based immunodiagnosis of tuberculosis: com-parison between whole-blood and enzyme-linked immunospot methods. J ClinMicrobiol. 2004;42:829-31. [PMID: 14766863]A22. Liebeschuetz S, Bamber S, Ewer K, Deeks J, Pathan AA, Lalvani A.Diagnosis of tuberculosis in South African children with a T-cell-based assay: aprospective cohort study. Lancet. 2004;364:2196-203. [PMID: 15610806]A23. Nicol MP, Pienaar D, Wood K, Eley B, Wilkinson RJ, Henderson H, etal. Enzyme-linked immunospot assay responses to early secretory antigenic target6, culture filtrate protein 10, and purified protein derivative among children withtuberculosis: implications for diagnosis and monitoring of therapy. Clin InfectDis. 2005;40:1301-8. [PMID: 15825033]A24. Aiken AM, Hill PC, Fox A, McAdam KP, Jackson-Sillah D, Lugos MD,et al. Reversion of the ELISPOT test after treatment in Gambian tuberculosiscases. BMC Infect Dis. 2006;6:66. [PMID: 16573826]A25. Meier T, Eulenbruch HP, Wrighton-Smith P, Enders G, Regnath T.Sensitivity of a new commercial enzyme-linked immunospot assay (T SPOT-TB)for diagnosis of tuberculosis in clinical practice. Eur J Clin Microbiol Infect Dis.2005;24:529-36. [PMID: 16133410]A26. Rolinck-Werninghaus C, Magdorf K, Stark K, Lyashchenko K, GennaroML, Colangeli R, et al. The potential of recombinant antigens ESAT-6, MPT63and mig for specific discrimination of Mycobacterium tuberculosis and M. aviuminfection [Letter]. Eur J Pediatr. 2003;162:534-6. [PMID: 12715165]A27. Taggart EW, Hill HR, Ruegner RG, Litwin CM. Evaluation of an in vitroassay for interferon gamma production in response to the Mycobacterium tuber-culosis-synthesized peptide antigens ESAT-6 and CFP-10 and the PPD skin test.Am J Clin Pathol. 2006;125:467-73. [PMID: 16613353]A28. Carrara S, Vincenti D, Petrosillo N, Amicosante M, Girardi E, Goletti D.Use of a T cell-based assay for monitoring efficacy of antituberculosis therapy.Clin Infect Dis. 2004;38:754-6. [PMID: 14986262]A29. Wilkinson KA, Kon OM, Newton SM, Meintjes G, Davidson RN, Pas-vol G, et al. Effect of treatment of latent tuberculosis infection on the T cellresponse to Mycobacterium tuberculosis antigens. J Infect Dis. 2006;193:354-9.[PMID: 16388482]A30. Ewer K, Millington KA, Deeks JJ, Alvarez L, Bryant G, Lalvani A. Dy-namic antigen-specific T-cell responses after point-source exposure to Mycobacte-rium tuberculosis. Am J Respir Crit Care Med. 2006;174:831-9. [PMID:16799072]A31. Pai M, Joshi R, Dogra S, Mendiratta DK, Narang P, Dheda K, et al.Persistently elevated T cell interferon-gamma responses after treatment for latent
Annals of Internal Medicine
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 W-81
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
tuberculosis infection among health care workers in India: a preliminary report. JOccup Med Toxicol. 2006;1:7. [PMID: 16722616]A32. Lalvani A, Pathan AA, Durkan H, Wilkinson KA, Whelan A, Deeks JJ, etal. Enhanced contact tracing and spatial tracking of Mycobacterium tuberculosisinfection by enumeration of antigen-specific T cells. Lancet. 2001;357:2017-21.[PMID: 11438135]A33. Ewer K, Deeks J, Alvarez L, Bryant G, Waller S, Andersen P, et al.Comparison of T-cell-based assay with tuberculin skin test for diagnosis of My-cobacterium tuberculosis infection in a school tuberculosis outbreak. Lancet. 2003;361:1168-73. [PMID: 12686038]A34. Richeldi L, Ewer K, Losi M, Bergamini BM, Roversi P, Deeks J, et al.T cell-based tracking of multidrug resistant tuberculosis infection after brief ex-posure. Am J Respir Crit Care Med. 2004;170:288-95. [PMID: 15130907]A35. Hill PC, Brookes RH, Fox A, Fielding K, Jeffries DJ, Jackson-Sillah D, etal. Large-scale evaluation of enzyme-linked immunospot assay and skin test fordiagnosis of Mycobacterium tuberculosis infection against a gradient of exposure inThe Gambia. Clin Infect Dis. 2004;38:966-73. [PMID: 15034828]A36. Zellweger JP, Zellweger A, Ansermet S, de Senarclens B, Wrighton-SmithP. Contact tracing using a new T-cell-based test: better correlation with tubercu-losis exposure than the tuberculin skin test. Int J Tuberc Lung Dis. 2005;9:1242-7. [PMID: 16333932]A37. Shams H, Weis SE, Klucar P, Lalvani A, Moonan PK, Pogoda JM, et al.Enzyme-linked immunospot and tuberculin skin testing to detect latent tubercu-losis infection. Am J Respir Crit Care Med. 2005;172:1161-8. [PMID:16081545]A38. Hill PC, Brookes RH, Adetifa IM, Fox A, Jackson-Sillah D, Lugos MD,et al. Comparison of enzyme-linked immunospot assay and tuberculin skin testin healthy children exposed to Mycobacterium tuberculosis. Pediatrics. 2006;117:1542-8. [PMID: 16651307]A39. Brock I, Weldingh K, Lillebaek T, Follmann F, Andersen P. Comparison
of tuberculin skin test and new specific blood test in tuberculosis contacts. Am JRespir Crit Care Med. 2004;170:65-9. [PMID: 15087297]A40. Nakaoka H, Lawson L, Squire SB, Coulter B, Ravn P, Brock I, et al. Riskfor tuberculosis among children. Emerg Infect Dis. 2006;12:1383-8. [PMID:17073087]A41. Diel R, Nienhaus A, Lange C, Meywald-Walter K, Forssbohm M,Schaberg T. Tuberculosis contact investigation with a new, specific blood test ina low-incidence population containing a high proportion of BCG-vaccinatedpersons. Respir Res. 2006;7:77. [PMID: 16707012]A42. Porsa E, Cheng L, Seale MM, Delclos GL, Ma X, Reich R, et al. Com-parison of a new ESAT-6/CFP-10 peptide-based gamma interferon assay and atuberculin skin test for tuberculosis screening in a moderate-risk population. ClinVaccine Immunol. 2006;13:53-8. [PMID: 16426000]A43. Pai M, Gokhale K, Joshi R, Dogra S, Kalantri S, Mendiratta DK, et al.Mycobacterium tuberculosis infection in health care workers in rural India: com-parison of a whole-blood interferon gamma assay with tuberculin skin testing.JAMA. 2005;293:2746-55. [PMID: 15941804]A44.Tsiouris SJ, Austin J, Toro P, Coetzee D, Weyer K, Stein Z, etal. Results of a tuberculosis-specific IFN-gamma assay in children at high riskfor tuberculosis infection. Int J Tuberc Lung Dis. 2006;10:939-41. [PMID:16898381]A45. Mahomed H, Hughes EJ, Hawkridge T, Minnies D, Simon E, Little F, etal. Comparison of mantoux skin test with three generations of a whole bloodIFN-gamma assay for tuberculosis infection. Int J Tuberc Lung Dis. 2006;10:310-6. [PMID: 16562712]A46. Harada N, Nakajima Y, Higuchi K, Sekiya Y, Rothel J, Mori T. Screeningfor tuberculosis infection using whole-blood interferon-gamma and Mantouxtesting among Japanese healthcare workers. Infect Control Hosp Epidemiol.2006;27:442-8. [PMID: 16671023]
W-82 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Appendix Table 1. Characteristics of the 3 Tests for Latent Tuberculosis Infection*
Variable Tuberculin Skin Test QuantiFERON-Gold andQuantiFERON-GoldIn-Tube
Elispot T-SPOT.TB
Administration In vivo (intradermal) Ex vivo, ELISA-based Ex vivo, Elispot-basedAntigens PPDS or RT-23 ESAT-6, CFP-10, and TB7.7 ESAT-6 and CFP-10Standardized Mostly Yes YesUnits of measurement Millimeters of
indurationUnits of IFN-� IFN-� spot-forming cells
Definition of positive test results 5, 10, and 15 mm Patient’s IFN-� �0.35U/mL (after subtractingIFN-� response in nilcontrol)
�6 spot-forming cells inthe antigen wells, with250 000 cells/well,and at least double-negative well
Indeterminate If anergy (rarely tested) Poor response to mitogen(�0.5 U/mL in positivecontrol) or highbackground response(�8.0 U/mL in nil well)
Poor response tomitogen (�20spot-forming cells inpositive control well)or high background(�10 spot-formingcells in negative well)
Time to result 48–72 h 16–24 h (but longer if runin batches)
* ELISA � enzyme-linked immunosorbent assay; IFN � interferon; IGRA � interferon-� release assay; PPDS � purified protein derivative standard.† All costs are in Canadian dollars ($1 Canadian � $0.91 U.S.). The numbers in parentheses are references. For the IGRA tests, the materials cost is based on quotes fromthe manufacturers for shipment to Canadian centers in September 2006. The cost for IGRA labor, shipping, and handling is taken from published field experience withQuantiFERON testing, as reported in reference A1. Costs may vary widely in different countries.
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 W-83
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Appendix Table 2. Summary of Methodological Aspects of Studies Included in the Review*
Study, Year(Reference)
Country Age,y†
DiagnosticCriteria
IntervalfromStart ofTBTherapyuntilIGRA, d
Blinding‡ Results on QFT or Elispot Results on TST
* Active tuberculosis was used as a surrogate marker for latent tuberculosis infection. C � clinical; H � histologic; IGRA � interferon-� release assay; M � microbiologic;NS � not stated; QFT-Q � QuantiFERON-Gold; TB � tuberculosis; TST � tuberculin skin test.† If given, the median or mean age is shown.‡ If yes, the technicians performing the IGRAs were blinded to the clinical condition of the patients. If no, the technicians were aware of the clinical condition of the patients.§ A 5-mm definition was used when the results were available, and a 10-mm definition was used if only those results were given.� 3TU-PPD was used instead of 5TU-PPD.¶ The Heaf test with 1TU-RT23 was used.
W-84 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Appendix Table 3. Specificity of QuantiFERON, Elispot, and Tuberculin Skin Test (TST) in Populations at Very Low Risk for LatentTuberculosis Infection*
Study, Year(Reference)
Country Age,y†
PopulationRiskAscertained‡
Blinding§ BCGVerifi-cation�
BCGStatus
QFT or Elispot TST¶
Positive,n
Total,n
Specificity,%
Cut-point,mm
Positive,n
Total,n
Specificity,%
QFT (ESAT-6only)
Johnson et al.,1999 (A4)
Australia 20 C NS D No BCGBCG
00
6054
100100
1015
01
6054
10098
QFT (ESAT-6andCFP-10)
Brock et al.,2001 (A5)
Denmark 3526
CC
NS NSNS
BCGNo BCG
00
1915
100100
– – – –
Mori et al.,2004 (A6)
Japan 20 C NS H BCG 4 213 98 10 73 113 36
Kang et al.,2005 (A7)
Korea 25 C Yes Scar BCG 4 99 96 10 50 99 49
Ravn et al.,2004 (A8)
Denmark 37 I NS H BCG 1 39 97 – – – –
Taggart et al.,2006 (A27)
UnitedStates
38 C NS H No BCG 0 81 100 15 3 81 96¶
Lee et al.,2006 (A12)
Korea 15 C NS H BCG 11 131 92 10 28 131 79
Elispot (ESAT-6only)
Lalvani et al.,2001 (A17)
UnitedKingdom
Adults I NS H, S BCG 0 26 100 – – – –
Lalvani et al.,2001 (A19)
UnitedKingdom
32 C No – BCG 0 40 100 – – – –
Pathan et al.,2001 (A18)
UnitedKingdom
32 I NS H BCG 0 32 100 – – – –
Elispot (ESAT-6andCFP-10)
Lee et al.,2006 (A12)
Korea 15 C NS H BCG 18 131 86 10 28 131 79
* BCG � bacille Calmette-Guerin; NS � not stated.† If given, the median or mean age is shown.‡ Risk ascertainment: C � complete (all factors questioned); I � incomplete (only some factors ascertained).§ If yes, the technicians performing the interferon-� release assays were blinded to the clinical condition of the patients. If no, the technicians performing the assay were awareof the clinical condition of the patients.� D � documented; H � history; scar � the patient was examined for a scar.¶ TST cut-point: A 10-mm definition was used for participants at low risk, but a 15 mm definition is used if only those results were given.
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 W-85
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Appendix Table 4. Results of Serial Testing for Patients Receiving Treatment*
2001 (A18)United Kingdom 12 ATB M, C 33 (NS) NS Not provided
Carrara et al.,2004 (A28)
Italy 18 ATB M 100 (100) NS Yes
Nicol et al.,2005 (A23)
South Africa 70 ATB M, C 60� (23) NS Not provided
Aiken et al.,2006 (A24)
Gambia 89 ATB M 59 (100) NS Yes
Wilkinson et al.,2006 (A29)
United Kingdom 33 LTBI TST- and IGRA-positive 48 (100) Yes NA
Ewer et al.,2006 (A30)
United Kingdom 63 LTBI TST- and IGRA-positive 35 (100) NS NA
QuantiFERONPai et al.,
2006 (A31)India 10 LTBI TST- and IGRA-positive 63 (100) Yes NA
* The intervals between testing were clearly stated in all studies. ATB � active tuberculosis; C � clinical diagnosis; IGRA � interferon-� release assay; LTBI � latenttuberculosis infection; M � microbiologic confirmation; NA � not applicable; NS � not stated; TST � tuberculin skin test.† The value in parentheses is the percentage of participants who completed all assays.‡ Adherence was assessed during treatment.§ If yes, treated outcomes were analyzed and provided.� In this study, of all eligible participants, 60% were retested at 1 month, 36% were retested at 3 months, and 14% were retested at 6 months.
Appendix Table 5. Studies That Estimated Sensitivity among Contacts for Whom a Gradient of Exposure was Defined*
Study, Year (Reference) Country ExposureDefineda Priori
Interval from TST to IGRA Blinding
TST IGRA
ElispotLalvani et al., 2001 (A32) United Kingdom Yes Same day No NSEwer et al., 2003 (A33) United Kingdom Yes 2–3 m NS YesRicheldi et al., 2004 (A34) Italy Yes Same day Yes NSHill et al., 2004 (A35) Gambia Yes Same day NS YesZellweger et al.,
2005 (A36)Switzerland NS† NS NS NS
Shams et al., 2005 (A37) United States NS† 0–14 d (60%) �14 d (33%) NS YesHill et al., 2006 (A38) Gambia Yes Same day NS Yes
QuantiFERONBrock et al., 2004 (A39) Denmark Yes 3 d after TST No NSKang et al., 2005 (A7) Korea Yes Same day Yes YesNakaoka et al.,
2006 (A40)Nigeria Yes Same day Yes NS
* IGRA � interferon-� release assay; NS � not stated; TST � tuberculin skin test.† Exposure scores were calculated on the basis of type, closeness, and duration of contact. In all other studies, exposure was defined as a single criterion: smear-positive orsmear-negative index cases, and same house or room versus different house.
W-86 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 www.annals.org
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Appendix Table 6. Studies That Examined Concordance of Interferon-� Release Assay (IGRA) and Tuberculin Skin Test (TST)*
Study, Year (Reference) Country Patients,n
BCGVerification†
TST andIGRASameDay‡
Blinding§
TST IGRA
No BCG vaccinationBrock et al., 2004 (A39) Denmark 45 H Yes NS NSDiel et al., 2006 (A41) Germany 309 H, scar Yes Yes YesPorsa et al., 2006 (A42) United States 409 H Yes NS Yes
BCG vaccination in infancyPai et al., 2005 (A43) India 719 H Yes NS NSDogra et al., 2006 (A14) India 97 Scar Yes Yes NSNakaoka et al., 2006 (A40) Nigeria 207 Scar Yes NS YesTsiouris et al., 2006 (A44) South Africa 184 Scar Yes Yes NSMahomed et al.,
2006 (A45)South Africa 358 Scar Yes NS NS
BCG vaccination afterinfancy (or varying orunspecified age ofvaccination)
Kang et al., 2005 (A7) Korea 120 Scar Yes Yes YesLee et al., 2006 (A12) Korea 131 H Yes NS NSConnell et al., 2006 (A10) Australia 98 Scar NS NS NSFerrara et al., 2006 (A11) Italy 286 Scar NS NS YesHarada et al., 2006 (A46) Japan 304 H Yes Yes NSDiel et al., 2006 (A41) Germany 309 H, scar Yes Yes Yes
* BCG � bacille Calmette–Guerin; H � history; NS � not stated; TST � tuberculin skin test.† The participants were examined for the presence of a scar.‡ If yes, the 2 tests were performed on the same day.§ If yes, the technicians performing the tests were blinded to the clinical condition of the patients. If no, the technicians performing the tests were aware of the clinicalcondition of the patients.
www.annals.org 6 March 2007 Annals of Internal Medicine Volume 146 • Number 5 W-87
Downloaded From: http://annals.org/ by a Oakland University User on 08/08/2013
Appendix Figure. Summary of literature search and study selection.*
Identified fromPubMed literaturesearch (n = 275)
Selected titles(n = 83)
Total full-textarticles reviewed
(n = 110)
Added from referencesand reviews
(n = 21)
Added in finalupdate in October 2006
(n = 6)
Included in review (n = 58)Sensitivity:
Active TB: 22Exposure gradient: 10
Specificity: 10Concordance of tests: 22Effect of immune suppression: 5Pediatric patients: 7Effect of treatment: 7*
